1. Field of the Technology
The disclosure relates to the field of multi-axis MEMS inertial measurement units (IMU) and in particular to units that exhibit temperature and vibration immunity, which are inherently quasi-digital, have ultra-high precision, stability, and wide dynamic range and wide measurement bandwidth.
2. Description of the Prior Art
All inertial measurement units (IMUs) are based on a combination of inertial sensors such as gyroscopes and accelerometers. The performance parameters of an IMU are defined by the parameters of the individual inertial sensors, as well as the method of processing the individual measurements together. Current MEMS IMUs do not meet the navigation grade requirements, and very few systems are reaching the tactical grade due to limitations in temperature and vibration sensitivity, signal stability, resolution, dynamic range, and bandwidth.
Conventional methods to create a compact MEMS IMU fail into three general regimes. The most common approach is to use off-the-shelf single-axis sensors mounted onto printed circuit boards (PCBs) and assembled into a three dimensional configuration. Single-axis sensors optimized to reject off-axis inputs are typically used, benefiting the overall IMU performance compared to using multi-axis sensors with a single proof mass. The approach is also well supported by the mature PCB technology. This approach does not allow further miniaturization and cost reduction since individual dies have to be fabricated, packaged, and tested separately. Also, conventional sensors based on analog amplitude modulation do not provide the necessary stability and immunity to vibration and temperature.
Another emerging but conventional method for creating micro IMUs involves chip-stacking. Each sensor is fabricated independently and then known-good dies are stacked together onto a single chip. In comparison to the PCB and common substrate approaches, the overall size of the IMU is reduced to a smaller footprint. However, for both methods alignment errors will vary for each device due to temperature, ambient vibrations, shock, accelerations, life span of IMU, limiting the effectiveness of factory calibration. This method also suffers from the inherent limitations of the conventional MEMS sensors operating based on analog amplitude modulation of signals.
A third alternative conventional method used to create a chip-level IMU is to fabricate all sensors onto a single die. This allows the footprint of the system as a whole to be small enough for chip-level packaging. Despite these advantages, there are significant drawbacks to this IMU architecture. First, creating conventional gyroscope(s) and accelerometer(s) requires very different packaging parameters. High precision gyroscope required high vacuum for operation; at the same time, accelerometers operate at atmospheric or above atmospheric pressure to provide the necessary bandwidth. This fundamental difference between packaging of high performance MEMS gyroscopes and accelerometers makes true co-fabrication and co-packaging impossible, preventing true integration.
All commercially available silicon MEMS gyroscopes and accelerometers rely on amplitude modulation of the input stimulus, where the inertial input produces a proportional change in the sensor output voltage. In other words, the inertial input is amplitude modulated. In this approach, the final output signal of a sensor is proportional to the true input, as well as a number of device parameters, including the stiffness of the spring, pick-up electronics gain, and so on. These additional contributors to the sensor bias and scale factor require calibration of each individual sensor. Variation of these internal parameters with time and with varying environment produces unpredictable drifts in the sensor output.
Another inherent disadvantage of conventional MEMS sensors using amplitude modulated signals comes from the limited dynamic range (the ratio between the full scale linear range and the smallest useful signal). In the best case scenario, AM capacitive readout with preselected low-noise electronic components can only achieve a resolution of 1×10−6 of the full scale, with a practical limit of 1×10−5. This means that achieving a 106 dynamic range and 1 ppm stability (requirement of the navigation grade) is practically impossible with conventional MEMS sensors. These fundamental limitations on the dynamic range and output stability prevent the use of MEMS gyroscopes and accelerometers in many important applications. AM architecture of sensors also presents tradeoffs between sensor noise and bandwidth (classic gain vs. bandwidth tradeoff of analog AM systems). The conventional AM based inertial sensor operation is also sensitive to temperature, resulting in significant response drifts over ambient temperature and pressure variations. The stability of the analogue signal reference is limited by 10−6, that fact creates irresolvable limitations on the AM devices stability.
For gyroscopes in particular, mode matching of high-Q angular rate gyroscopes increases the signal-to noise-ratio at the tradeoff of linear range and measurement bandwidth (10 deg/s range, sub-Hz bandwidth typical for Q˜100 k). These constraints stem from a fundamental Q versus bandwidth tradeoff and dynamic range limitations of analog amplitude modulation (AM) systems (in conventional MEMS gyroscopes, the sense-mode response is excited by the input angular rate amplitude-modulated by the drive-mode velocity). The conventional AM based gyroscopes operation is also sensitive to the value of the sense-mode Q-factor, resulting in significant response drifts over ambient temperature and pressure variations.
At the same time, the conventional AM type accelerometer must have very low Q-factor (single digit values—100,000 less than Q of a gyroscope) to provide bandwidth, which requires encapsulating it in air or high pressure. The fact that conventional MEMS gyroscopes are inherently resonant devices while conventional MEMS accelerometers are inherently non-resonant devices makes it impossible to truly integrate them together. As a result conventional wide bandwidth, high precision accelerometers and gyroscopes are fabricated separately.
In response to the foregoing problems a frequency modulated accelerometer has been proposed in the art, where induced acceleration changes the resonant frequency of the device due to changes in the total effective stiffness (combination of positive mechanical stiffness and negative electrostatic spring). The main challenge to overcome in this approach is temperature sensitivity of the resonant frequency, caused by the dependency of the silicon's Young's modulus on temperature.